OpenClaw / Hermes Agent secure remote setup

OpenClaw (formerly Clawdbot / Moltbot) and Nous Research's Hermes Agent are open-source, self-hosted AI agents that need a server to stay online 24/7. We install them, get them running, and keep the admin dashboard off the public internet.

Plans

Need both projects, more channels or more models? Email us and we'll quote.

What basic hardening covers

• Dashboard not exposed to the public internet: the dashboard and gateway listen on localhost and you reach them through an SSH tunnel. If you really need public access, we restrict source IPs in your firewall first. OpenClaw's security docs note that a regular host install binds the Gateway to loopback, while container images default to an exposed bind that must be paired with Gateway auth.
• Access password: strong random passwords for the dashboard login and gateway auth, stored in an admin-only config file. We show you where to view and change them.
• Least privilege: only required ports are opened, services don't run with more privileges than they need, and model API keys live only in the service config.
• Backup: a backup of the config and data directories before handover, plus instructions to back up and restore yourself.

Channels available in Advanced

We only connect channels listed in each project's official documentation, for example:

• OpenClaw: Telegram, Slack, Discord, WhatsApp, Microsoft Teams, Signal, Matrix, Feishu and others in the official channel catalog (WeCom is available as an external plugin).
• Hermes Agent: Telegram, Discord, Slack, WhatsApp, Signal, Matrix, Email, Feishu/Lark, DingTalk, WeCom and others in the official Messaging docs.

You create the bot, app or business account on the messaging platform yourself under its rules; we configure it following the official steps.

Model API keys

You open and pay for your model API key with the provider yourself. We don't buy, resell or share API keys.

How it works

1. Contact us at agentcommx@gmail.com: OpenClaw or Hermes Agent, which plan, and which channel and model you want.
2. Confirm requirements and server spec.
3. You grant temporary access: a temporary SSH user with root or sudo, or a temporary password, created by you.
4. Deployment and hardening by our engineering team, plus model and channel setup for Advanced.
5. Acceptance check: you try it from the dashboard or your chat channel.
6. You revoke access by changing the password or deleting the temporary user.
7. 7 days of Q&A, counted from the delivery date.

Delivery time

Usually within 3 business days after the requirements are confirmed and access has been granted.

What you provide

• A Linux server you own (Ubuntu 22.04 / 24.04 or Debian 12 recommended).
• Temporary root or sudo SSH access that you set up and later revoke yourself.
• For Advanced: your own model API key and the developer or business account on the messaging platform.
• A domain name is optional.

Access and automated tools

• Server access is used only for the agreed deployment. We don't keep your passwords, keys or API keys.
• Our engineering team may use automated tools, including AI assistants, to run and check the deployment steps, under the same temporary access and limits.

What we don't do

• No configurations that circumvent platform rules or regional restrictions, and no channels the projects don't officially support.
• We don't buy model credits, servers or channel accounts for you, and ongoing maintenance after the Q&A period isn't included.

Refunds and liability

Full refund before work starts, and a full refund if the deployment fails. Our liability for any order is limited to the amount paid for that order. See the Refund Policy and Terms of Service.

OpenClaw and Hermes Agent are MIT-licensed open-source projects. We are not affiliated with their authors or maintainers.

Last updated: 2026-10-18 · 中文